Topological Persistence Guided Knowledge Distillation for Wearable Sensor Data.

Deep learning methods have achieved a lot of success in various applications involving converting wearable sensor data to actionable health insights. A common application areas is activity recognition, where deep-learning methods still suffer from limitations such as sensitivity to signal quality, sensor characteristic variations, and variability between subjects. To mitigate these issues, robust features obtained by topological data analysis (TDA) have been suggested as a potential solution. However, there are two significant obstacles to using topological features in deep learning: (1) large computational load to extract topological features using TDA, and (2) different signal representations obtained from deep learning and TDA which makes fusion difficult. In this paper, to enable integration of the strengths of topological methods in deep-learning for time-series data, we propose to use two teacher networks - one trained on the raw time-series data, and another trained on persistence images generated by TDA methods. These two teachers are jointly used to distill a single student model, which utilizes only the raw time-series data at test-time. This approach addresses both issues. The use of KD with multiple teachers utilizes complementary information, and results in a compact model with strong supervisory features and an integrated richer representation. To assimilate desirable information from different modalities, we design new constraints, including orthogonality imposed on feature correlation maps for improving feature expressiveness and allowing the student to easily learn from the teacher. Also, we apply an annealing strategy in KD for fast saturation and better accommodation from different features, while the knowledge gap between the teachers and student is reduced. Finally, a robust student model is distilled, which can at test-time uses only the time-series data as an input, while implicitly preserving topological features. The experimental results demonstrate the effectiveness of the proposed method on wearable sensor data. The proposed method shows 71.74% in classification accuracy on GENEActiv with WRN16-1 (1D CNNs) student, which outperforms baselines and takes much less processing time (less than 17 sec) than teachers on 6k testing samples.

ROBUST TIME SERIES RECOVERY AND CLASSIFICATION USING TEST-TIME NOISE SIMULATOR NETWORKS.

Time-series are commonly susceptible to various types of corruption due to sensor-level changes and defects which can result in missing samples, sensor and quantization noise, unknown calibration, unknown phase shifts etc. These corruptions cannot be easily corrected as the noise model may be unknown at the time of deployment. This also results in the inability to employ pre-trained classifiers, trained on (clean) source data. In this paper, we present a general framework and models for time-series that can make use of (unlabeled) test samples to estimate the noise model-entirely at test time. To this end, we use a coupled decoder model and an additional neural network which acts as a learned noise model simulator. We show that the framework is able to "clean" the data so as to match the source training data statistics and the cleaned data can be directly used with a pre-trained classifier for robust predictions. We perform empirical studies on diverse application domains with different types of sensors, clearly demonstrating the effectiveness and generality of this method.

Role of Data Augmentation Strategies in Knowledge Distillation for Wearable Sensor Data.

Deep neural networks are parametrized by several thousands or millions of parameters, and have shown tremendous success in many classification problems. However, the large number of parameters makes it difficult to integrate these models into edge devices such as smartphones and wearable devices. To address this problem, knowledge distillation (KD) has been widely employed, that uses a pre-trained high capacity network to train a much smaller network, suitable for edge devices. In this paper, for the first time, we study the applicability and challenges of using KD for time-series data for wearable devices. Successful application of KD requires specific choices of data augmentation methods during training. However, it is not yet known if there exists a coherent strategy for choosing an augmentation approach during KD. In this paper, we report the results of a detailed study that compares and contrasts various common choices and some hybrid data augmentation strategies in KD based human activity analysis. Research in this area is often limited as there are not many comprehensive databases available in the public domain from wearable devices. Our study considers databases from small scale publicly available to one derived from a large scale interventional study into human activity and sedentary behavior. We find that the choice of data augmentation techniques during KD have a variable level of impact on end performance, and find that the optimal network choice as well as data augmentation strategies are specific to a dataset at hand. However, we also conclude with a general set of recommendations that can provide a strong baseline performance across databases.

Training calibration-based counterfactual explainers for deep learning models in medical image analysis.

The rapid adoption of artificial intelligence methods in healthcare is coupled with the critical need for techniques to rigorously introspect models and thereby ensure that they behave reliably. This has led to the design of explainable AI techniques that uncover the relationships between discernible data signatures and model predictions. In this context, counterfactual explanations that synthesize small, interpretable changes to a given query while producing desired changes in model predictions have become popular. This under-constrained, inverse problem is vulnerable to introducing irrelevant feature manipulations, particularly when the model's predictions are not well-calibrated. Hence, in this paper, we propose the TraCE (training calibration-based explainers) technique, which utilizes a novel uncertainty-based interval calibration strategy for reliably synthesizing counterfactuals. Given the wide-spread adoption of machine-learned solutions in radiology, our study focuses on deep models used for identifying anomalies in chest X-ray images. Using rigorous empirical studies, we demonstrate the superiority of TraCE explanations over several state-of-the-art baseline approaches, in terms of several widely adopted evaluation metrics. Our findings show that TraCE can be used to obtain a holistic understanding of deep models by enabling progressive exploration of decision boundaries, to detect shortcuts, and to infer relationships between patient attributes and disease severity.

Topological Knowledge Distillation for Wearable Sensor Data.

Converting wearable sensor data to actionable health insights has witnessed large interest in recent years. Deep learning methods have been utilized in and have achieved a lot of successes in various applications involving wearables fields. However, wearable sensor data has unique issues related to sensitivity and variability between subjects, and dependency on sampling-rate for analysis. To mitigate these issues, a different type of analysis using topological data analysis has shown promise as well. Topological data analysis (TDA) captures robust features, such as persistence images (PI), in complex data through the persistent homology algorithm, which holds the promise of boosting machine learning performance. However, because of the computational load required by TDA methods for large-scale data, integration and implementation has lagged behind. Further, many applications involving wearables require models to be compact enough to allow deployment on edge-devices. In this context, knowledge distillation (KD) has been widely applied to generate a small model (student model), using a pre-trained high-capacity network (teacher model). In this paper, we propose a new KD strategy using two teacher models - one that uses the raw time-series and another that uses persistence images from the time-series. These two teachers then train a student using KD. In essence, the student learns from heterogeneous teachers providing different knowledge. To consider different properties in features from teachers, we apply an annealing strategy and adaptive temperature in KD. Finally, a robust student model is distilled, which utilizes the time series data only. We find that incorporation of persistence features via second teacher leads to significantly improved performance. This approach provides a unique way of fusing deep-learning with topological features to develop effective models.

Digital medicine and the curse of dimensionality.

Digital health data are multimodal and high-dimensional. A patient's health state can be characterized by a multitude of signals including medical imaging, clinical variables, genome sequencing, conversations between clinicians and patients, and continuous signals from wearables, among others. This high volume, personalized data stream aggregated over patients' lives has spurred interest in developing new artificial intelligence (AI) models for higher-precision diagnosis, prognosis, and tracking. While the promise of these algorithms is undeniable, their dissemination and adoption have been slow, owing partially to unpredictable AI model performance once deployed in the real world. We posit that one of the rate-limiting factors in developing algorithms that generalize to real-world scenarios is the very attribute that makes the data exciting-their high-dimensional nature. This paper considers how the large number of features in vast digital health data can challenge the development of robust AI models-a phenomenon known as "the curse of dimensionality" in statistical learning theory. We provide an overview of the curse of dimensionality in the context of digital health, demonstrate how it can negatively impact out-of-sample performance, and highlight important considerations for researchers and algorithm designers.

Automated Movement Assessment in Stroke Rehabilitation.

We are developing a system for long term Semi-Automated Rehabilitation At the Home (SARAH) that relies on low-cost and unobtrusive video-based sensing. We present a cyber-human methodology used by the SARAH system for automated assessment of upper extremity stroke rehabilitation at the home. We propose a hierarchical model for automatically segmenting stroke survivor's movements and generating training task performance assessment scores during rehabilitation. The hierarchical model fuses expert therapist knowledge-based approaches with data-driven techniques. The expert knowledge is more observable in the higher layers of the hierarchy (task and segment) and therefore more accessible to algorithms incorporating high level constraints relating to activity structure (i.e., type and order of segments per task). We utilize an HMM and a Decision Tree model to connect these high level priors to data driven analysis. The lower layers (RGB images and raw kinematics) need to be addressed primarily through data driven techniques. We use a transformer based architecture operating on low-level action features (tracking of individual body joints and objects) and a Multi-Stage Temporal Convolutional Network(MS-TCN) operating on raw RGB images. We develop a sequence combining these complimentary algorithms effectively, thus encoding the information from different layers of the movement hierarchy. Through this combination, we produce a robust segmentation and task assessment results on noisy, variable and limited data, which is characteristic of low cost video capture of rehabilitation at the home. Our proposed approach achieves 85% accuracy in per-frame labeling, 99% accuracy in segment classification and 93% accuracy in task completion assessment. Although the methodology proposed in this paper applies to upper extremity rehabilitation using the SARAH system, it can potentially be used, with minor alterations, to assist automation in many other movement rehabilitation contexts (i.e., lower extremity training for neurological accidents).

Unsupervised Pre-trained Models from Healthy ADLs Improve Parkinson's Disease Classification of Gait Patterns.

Application and use of deep learning algorithms for different healthcare applications is gaining interest at a steady pace. However, use of such algorithms can prove to be challenging as they require large amounts of training data that capture different possible variations. This makes it difficult to use them in a clinical setting since in most health applications researchers often have to work with limited data. Less data can cause the deep learning model to over-fit. In this paper, we ask how can we use data from a different environment, different use-case, with widely differing data distributions. We exemplify this use case by using single-sensor accelerometer data from healthy subjects performing activities of daily living - ADLs (source dataset), to extract features relevant to multi-sensor accelerometer gait data (target dataset) for Parkinson's disease classification. We train the pre-trained model using the source dataset and use it as a feature extractor. We show that the features extracted for the target dataset can be used to train an effective classification model. Our pretrained source model consists of a convolutional autoencoder, and the target classification model is a simple multi-layer perceptron model. We explore two different pre-trained source models, trained using different activity groups, and analyze the influence the choice of pre-trained model has over the task of Parkinson's disease classification.

Topological Descriptors for Parkinson's Disease Classification and Regression Analysis.

At present, the vast majority of human subjects with neurological disease are still diagnosed through in-person assessments and qualitative analysis of patient data. In this paper, we propose to use Topological Data Analysis (TDA) together with machine learning tools to automate the process of Parkinson's disease classification and severity assessment. An automated, stable, and accurate method to evaluate Parkinson's would be significant in streamlining diagnoses of patients and providing families more time for corrective measures. We propose a methodology which incorporates TDA into analyzing Parkinson's disease postural shifts data through the representation of persistence images. Studying the topology of a system has proven to be invariant to small changes in data and has been shown to perform well in discrimination tasks. The contributions of the paper are twofold. We propose a method to 1) classify healthy patients from those afflicted by disease and 2) diagnose the severity of disease. We explore the use of the proposed method in an application involving a Parkinson's disease dataset comprised of healthy-elderly, healthy-young and Parkinson's disease patients. Our code is available at https://github.com/itsmeafra/Sublevel-Set-TDA.

PI-Net: A Deep Learning Approach to Extract Topological Persistence Images.

Topological features such as persistence diagrams and their functional approximations like persistence images (PIs) have been showing substantial promise for machine learning and computer vision applications. This is greatly attributed to the robustness topological representations provide against different types of physical nuisance variables seen in real-world data, such as view-point, illumination, and more. However, key bottlenecks to their large scale adoption are computational expenditure and difficulty incorporating them in a differentiable architecture. We take an important step in this paper to mitigate these bottlenecks by proposing a novel one-step approach to generate PIs directly from the input data. We design two separate convolutional neural network architectures, one designed to take in multi-variate time series signals as input and another that accepts multi-channel images as input. We call these networks Signal PI-Net and Image PI-Net respectively. To the best of our knowledge, we are the first to propose the use of deep learning for computing topological features directly from data. We explore the use of the proposed PI-Net architectures on two applications: human activity recognition using tri-axial accelerometer sensor data and image classification. We demonstrate the ease of fusion of PIs in supervised deep learning architectures and speed up of several orders of magnitude for extracting PIs from data. Our code is available at https://github.com/anirudhsom/PI-Net.

A Framework for Interpretable Full-Body Kinematic Description Using Geometric and Functional Analysis.

Rapid advances in cost-effective and non-invasive depth sensors, and the development of reliable and real-time 3D skeletal data estimation algorithms, have opened up a new application area in computer vision - statistical analysis of human kinematic data for fast, automated assessment of body movements. These assessments can play important roles in sports, medical diagnosis, physical therapy, elderly monitoring and related applications. This paper develops a comprehensive geometric framework for quantification and statistical evaluation of kinematic features. The key idea is to avoid analysis of individual joints, as is the current paradigm, and represent movements as temporal evolutions, or trajectories, on shape space of full body skeletons. This allows metrics with appropriate invariance properties to be imposed on these trajectories and leads to definitions of higher-level features, such as spatial symmetry (sS), temporal symmetry (tS), action's velocity (Vl) and body's balance (Bl), during performance of an action. These features exploit skeletal symmetries in space and time, and capture motion cadence to naturally quantify motions of individual subjects. The study of these features as functional data allows us to formulate certain hypothesis tests in feature space. This, in turn, leads to validation of existing assumptions and discoveries of new relationships between kinematics and demographic factors, such as age, gender, and athletic training. We use the clinically validated K3Da kinect dataset to illustrate these ideas, and hope these tools will lead to discovery of new relationships between full-body kinematic features and demographic, health, and wellness factors that are clinically relevant.

An Optical Flow-Based Approach for Minimally Divergent Velocimetry Data Interpolation.

Three-dimensional (3D) biomedical image sets are often acquired with in-plane pixel spacings that are far less than the out-of-plane spacings between images. The resultant anisotropy, which can be detrimental in many applications, can be decreased using image interpolation. Optical flow and/or other registration-based interpolators have proven useful in such interpolation roles in the past. When acquired images are comprised of signals that describe the flow velocity of fluids, additional information is available to guide the interpolation process. In this paper, we present an optical-flow based framework for image interpolation that also minimizes resultant divergence in the interpolated data.

Elastic Functional Coding of Riemannian Trajectories.

Visual observations of dynamic phenomena, such as human actions, are often represented as sequences of smoothly-varying features. In cases where the feature spaces can be structured as Riemannian manifolds, the corresponding representations become trajectories on manifolds. Analysis of these trajectories is challenging due to non-linearity of underlying spaces and high-dimensionality of trajectories. In vision problems, given the nature of physical systems involved, these phenomena are better characterized on a low-dimensional manifold compared to the space of Riemannian trajectories. For instance, if one does not impose physical constraints of the human body, in data involving human action analysis, the resulting representation space will have highly redundant features. Learning an effective, low-dimensional embedding for action representations will have a huge impact in the areas of search and retrieval, visualization, learning, and recognition. Traditional manifold learning addresses this problem for static points in the euclidean space, but its extension to Riemannian trajectories is non-trivial and remains unexplored. The difficulty lies in inherent non-linearity of the domain and temporal variability of actions that can distort any traditional metric between trajectories. To overcome these issues, we use the framework based on transported square-root velocity fields (TSRVF); this framework has several desirable properties, including a rate-invariant metric and vector space representations. We propose to learn an embedding such that each action trajectory is mapped to a single point in a low-dimensional euclidean space, and the trajectories that differ only in temporal rates map to the same point. We utilize the TSRVF representation, and accompanying statistical summaries of Riemannian trajectories, to extend existing coding methods such as PCA, KSVD and Label Consistent KSVD to Riemannian trajectories or more generally to Riemannian functions. We show that such coding efficiently captures trajectories in applications such as action recognition, stroke rehabilitation, visual speech recognition, clustering and diverse sequence sampling. Using this framework, we obtain state-of-the-art recognition results, while reducing the dimensionality/ complexity by a factor of 100-250x. Since these mappings and codes are invertible, they can also be used to interactively-visualize Riemannian trajectories and synthesize actions.

Shape Distributions of Nonlinear Dynamical Systems for Video-Based Inference.

This paper presents a shape-theoretic framework for dynamical analysis of nonlinear dynamical systems which appear frequently in several video-based inference tasks. Traditional approaches to dynamical modeling have included linear and nonlinear methods with their respective drawbacks. A novel approach we propose is the use of descriptors of the shape of the dynamical attractor as a feature representation of nature of dynamics. The proposed framework has two main advantages over traditional approaches: a) representation of the dynamical system is derived directly from the observational data, without any inherent assumptions, and b) the proposed features show stability under different time-series lengths where traditional dynamical invariants fail. We illustrate our idea using nonlinear dynamical models such as Lorenz and Rossler systems, where our feature representations (shape distribution) support our hypothesis that the local shape of the reconstructed phase space can be used as a discriminative feature. Our experimental analyses on these models also indicate that the proposed framework show stability for different time-series lengths, which is useful when the available number of samples are small/variable. The specific applications of interest in this paper are: 1) activity recognition using motion capture and RGBD sensors, 2) activity quality assessment for applications in stroke rehabilitation, and 3) dynamical scene classification. We provide experimental validation through action and gesture recognition experiments on motion capture and Kinect datasets. In all these scenarios, we show experimental evidence of the favorable properties of the proposed representation.

Component-Level Tuning of Kinematic Features From Composite Therapist Impressions of Movement Quality.

In this paper, we propose a general framework for tuning component-level kinematic features using therapists' overall impressions of movement quality, in the context of a home-based adaptive mixed reality rehabilitation (HAMRR) system. We propose a linear combination of nonlinear kinematic features to model wrist movement, and propose an approach to learn feature thresholds and weights using high-level labels of overall movement quality provided by a therapist. The kinematic features are chosen such that they correlate with the quality of wrist movements to clinical assessment scores. Further, the proposed features are designed to be reliably extracted from an inexpensive and portable motion capture system using a single reflective marker on the wrist. Using a dataset collected from ten stroke survivors, we demonstrate that the framework can be reliably used for movement quality assessment in HAMRR systems. The system is currently being deployed for large-scale evaluations, and will represent an increasingly important application area of motion capture and activity analysis.

Attractor-shape descriptors for balance impairment assessment in Parkinson's disease.

In this paper, we propose a computational framework using high-dimensional shape descriptors of reconstructed attractors of center-of-pressure (CoP) tracings collected from subjects with Parkinson's disease while performing dynamical posture shifts, to quantitatively assess balance impairment. Using a dataset collected from 60 subjects, we demonstrated that the proposed method outperforms traditional methods, such as dynamical shift indices and use of chaotic invariants, in assessment of balance impairment.

A statistical estimation framework for energy expenditure of physical activities from a wrist-worn accelerometer.

Energy expenditure (EE) estimation from accelerometer-based wearable sensors is important to generate accurate assessment of physical activity (PA) in individuals. Approaches hitherto have mainly focused on using accelerometer data and features extracted from these data to learn a regression model to predict EE directly. In this paper, we propose a novel framework for EE estimation based on statistical estimation theory. Given a test sequence of accelerometer data, the probability distribution on the PA classes is estimated by a classifier and these predictions are used to estimate EE. Experimental evaluation, performed on a large dataset of 152 subjects and 12 activity classes, demonstrates that EE can be estimated accurately using our framework.

Reconstruction-Free Action Inference from Compressive Imagers.

Persistent surveillance from camera networks, such as at parking lots, UAVs, etc., often results in large amounts of video data, resulting in significant challenges for inference in terms of storage, communication and computation. Compressive cameras have emerged as a potential solution to deal with the data deluge issues in such applications. However, inference tasks such as action recognition require high quality features which implies reconstructing the original video data. Much work in compressive sensing (CS) theory is geared towards solving the reconstruction problem, where state-of-the-art methods are computationally intensive and provide low-quality results at high compression rates. Thus, reconstruction-free methods for inference are much desired. In this paper, we propose reconstruction-free methods for action recognition from compressive cameras at high compression ratios of 100 and above. Recognizing actions directly from CS measurements requires features which are mostly nonlinear and thus not easily applicable. This leads us to search for such properties that are preserved in compressive measurements. To this end, we propose the use of spatio-temporal smashed filters, which are compressive domain versions of pixel-domain matched filters. We conduct experiments on publicly available databases and show that one can obtain recognition rates that are comparable to the oracle method in uncompressed setup, even for high compression ratios.

Interdisciplinary concepts for design and implementation of mixed reality interactive neurorehabilitation systems for stroke.

Interactive neurorehabilitation (INR) systems provide therapy that can evaluate and deliver feedback on a patient's movement computationally. There are currently many approaches to INR design and implementation, without a clear indication of which methods to utilize best. This article presents key interactive computing, motor learning, and media arts concepts utilized by an interdisciplinary group to develop adaptive, mixed reality INR systems for upper extremity therapy of patients with stroke. Two INR systems are used as examples to show how the concepts can be applied within: (1) a small-scale INR clinical study that achieved integrated improvement of movement quality and functionality through continuously supervised therapy and (2) a pilot study that achieved improvement of clinical scores with minimal supervision. The notion is proposed that some of the successful approaches developed and tested within these systems can form the basis of a scalable design methodology for other INR systems. A coherent approach to INR design is needed to facilitate the use of the systems by physical therapists, increase the number of successful INR studies, and generate rich clinical data that can inform the development of best practices for use of INR in physical therapy.

Decision support for stroke rehabilitation therapy via describable attribute-based decision trees.

This paper proposes a computational framework for movement quality assessment using a decision tree model that can potentially assist a physical therapist in a telerehabilitation context. Using a dataset of key kinematic attributes collected from eight stroke survivors, we demonstrate that the framework can be reliably used for movement quality assessment of a reach-to-grasp cone task, an activity commonly used in upper extremity stroke rehabilitation therapy. The proposed framework is capable of providing movement quality scores that are highly correlated to the ratings provided by therapists, who used a custom rating rubric created by rehabilitation experts. Our hypothesis is that a decision tree model could be easily utilized by therapists as a potential assistive tool, especially in evaluating movement quality on a large-scale dataset collected during unsupervised rehabilitation (e.g., training at the home), thereby reducing the time and cost of rehabilitation treatment.